Case study 13 — Feedback triage automation
Customer Feedback Analyzer
Feedback usually dies in a form. Someone reads the inbox on Friday, decides what it was about, tells the right person, and maybe replies. Each of those steps is small, which is exactly why they get skipped.
- n8n
- Airtable
- Slack
- Gmail
- LLM prompting
The problem
Manual triage is slow and inconsistent. Feedback arrives, sits, gets categorised differently depending on who reads it, and reaches the responsible team late or not at all, while the customer hears nothing.
What we built
Six connected nodes: form submission, AI classification, routing logic, an Airtable record, a Slack notification and a Gmail follow-up. One workflow covers triage, routing and communication.
How it works
Classification before routing
The AI agent reads each submission and extracts what it is about and how urgent it is, so routing operates on meaning rather than on which form field someone filled in.
Routing as explicit rules
Rules and conditions send each item down the right internal path. The logic is inspectable rather than buried in a person's judgement, which is what makes it consistent between Monday and Friday.
A structured record, not an inbox
Every submission is written to Airtable with its name, contact details, feedback and status, so the history is queryable instead of scattered across a mailbox.
The team told in the channel they already watch
A Slack notification carries the name, email, feedback, category and priority into the relevant channel, so nobody has to open another tool to learn that something needs attention.
The customer acknowledged automatically
A Gmail follow-up sends the acknowledgement and any later update, closing the loop that manual triage most often leaves open.
In the product
Outcome
Manual triage is reduced, internal routing is immediate, customers get acknowledged without anyone remembering to do it, and feedback stays consistent across Airtable, Slack and email rather than living in one person's inbox.